PDA

View Full Version : Temporary Machine Freezes - Need Help Pleaseq



JBT27
04-23-2010, 06:37 AM
This is about the machine in my sig.

Simple really - after doing an F9, or an F10 sequence, or a RAM preview in AE, then want to go onto the desktop, or open a drive from a shortcut, or switch to another app, or actually anything, including copying, I have to wait anything between tens of seconds, and sometimes a minute or two .....

Now I know this is a length of a piece of string thing, but I'm doing standard def renders, and RAM previews, of 500 frames or so, nothing heavyweight, and yet the machine's got this thing of halting, then suddenly clearing and being available again.

You'd think paging file issues, but as I say, I don't reckon these are big things I'm doing to cause this.

Any ideas what can cause this?

Thanks.

Julian.

biliousfrog
04-23-2010, 06:55 AM
Probably worth running some diagnostics such as memtest and checkdisk and defragment your HD's. Also ensure that you haven't got any power save options enabled, maybe the HD's are spinning down?

JBT27
04-23-2010, 10:58 AM
Thanks.

Memtest is reporting errors - just the plain free version - got errors within seconds, though on an 8Gb machine I started running three copies set to check 2048 each. Then I couldn't shut one of them down, had to force quit it.

RAM intensive operations do seem to be causing some of the long pauses - RAM previewing in AE, rendering in LW .....

Great - right on top of a deadline.

Julian.

Tranimatronic
04-23-2010, 11:41 AM
8GB ? it that 1GB*8 / 2GB*4 / 4GB*2 ?
Hopefully it will be the second. If it is - remove two and try the memory tests again. If it persists, switch them and try again. This will cost nothing and may show you the culprit.
Good luck !

JBT27
04-23-2010, 11:49 AM
8GB ? it that 1GB*8 / 2GB*4 / 4GB*2 ?
Hopefully it will be the second. If it is - remove two and try the memory tests again. If it persists, switch them and try again. This will cost nothing and may show you the culprit.
Good luck !

It is the second!

Thanks ..... I was starting to wonder about that. On the proviso also that Memtest works fine with the RAM and is generally accurate .....

Julian.

JonW
04-23-2010, 03:21 PM
Try to use Bounding Boxes in your view ports where you can. If you Textures or Wire frames etc LW locks up that ram. You will need to change to BB & restart LW.

This also helps on my computer & it’s got 24gb of ram.

danielkaiser
04-23-2010, 03:25 PM
have you checked your event viewer.

Hopper
04-23-2010, 05:49 PM
It really does sound like a memory issue. To validate your findings with MemTest, you may want to go into your BIOS, and choose to do a full memory test on bootup (it's usually marked as 'slow'/'full', etc..). With 8GB of RAM, it will be quite painfully slow, but you should only really have to do it once. It will either report an error, or simply lockup while testing the offending module.

Unfortunately, it won't tell you which module is bad - only that it failed.

So as stated previously, try using only pairs (4GB) until you can pass the test. This obviously doesn't 'solve' your problem, but it will at least allow you to continue working while you get a replacement.

OnlineRender
04-24-2010, 03:48 AM
hmm does sound like a read and write cache error somewhere ,but....., just randomly hanging? , quick questions , does it hang when windows boots up ! for example ,it takes longer than it should for the OS to work fluent ? i.e hit windows key and it takes a few seconds to display ,even though your on a uber machine ?

the reason I'm asking is because I had the same problem when I upgraded AE , hung like a mother-f'er everytime i switched (alt+tab) between apps .

It could be mem error or even your HD starting to fail but my other suggestion would be to see if you have a dirty virus playing games with your syst3m .

even though it maybe an offline machine it still runs the high risk of being corrupt ."esp with usb sticks "

I found out that the virus settled in the bios "could have been worse , could easily go into cmos or physical memory " and it caused the memory to hang ! "as if the power failed to the stick " was a nightmare to remove ,I hope your lucky/cheaper if it is just a virus in your HD causing your c:\system file failure which would could result in your machine hanging also .

quick way to look for a virus if your'e system is online , go into DOS" netstat -a / -s "have a look for open connections , this should be a minimum amount ,also look to see if there is connections , that just should be there , most viruses written these days are designed to keep your system intact and open connections for the host . gone are the days that .bat .exe are dystroyed :D

JBT27
04-24-2010, 09:44 AM
Thanks for the replies.

Well, we've spent all morning with the guy who builds our machines, swapping out RAM modules, but got nowhere with that, ie. same performance no matter what.

We tried another graphics board, the self same one from the other quad-core we have, which is identical, but again same odd performance.

In fairness, I haven't run a full Memtest test, partly due to time constraints, but I reckoned that the sledgehammer approach of swapping out modules might have shown some differences or even improvement.

The lockups do not happen at startup at all, not once. Nor does the machine actually crash.

It is an offline machine. I'd say the vast majority of stuff that goes on this machine is generated here, and not from any external source; virtually nothing is derived from downloaded files or stuff off memory sticks and so on. That said, I'm not ruling out a virus either.

When I upgraded to AE CS4, middle of last year, none of this was an issue - worked great. However, what you are describing, OnlineRender, is pretty much what I'm getting now and have been for the past few weeks.

It's not exactly random. The lockups will happen either after a RAM preview in AE, rendering in LW, even a single F9, copying files from one drive to another or across the network. RAM previewing in AE is getting painfully slow to render - jitters and jerks forward mostly, but not always - plays fine once it's rendered the preview. This is new - these are standard def PAL frames, about 700 or so, using 2.3Gb RAM as showin in Task Manager while running. For the last four or five months of last year, I was running RAM previews of HD shots all day every day without this kind of issue ..... some slowdowns, as expected, but even then it was fairly quick to clear and return to normal performance.

Typically now, after doing one of the above, I'll try to select a menu in AE, or if I get onto the desktop click a drive shortcut, or the Start menu, but nothing will happen - the whole machine will be locked. Give it anything between a half and two minutes or so, and it will clear, and then mostly be OK, until it happens again.

Often when clicking a desktop icon, the icon will disappear, then the machine will lockup.

So, still trying to work through all the cleanups and virus checking I can. Not ruling out anything, but I genuinely thought the one of the RAM modules might have shown as flawed. That said, my gut sense is that it is some kind of read/write cache error, as OnlineRender suggests.

Julian.

OnlineRender
04-24-2010, 12:40 PM
meh either way , it's still a nightmare ! another quick question ,you never mentioned what OS it's running, "I gather WIN7 if your running that kinda stick " ? , i would dual boot , usually I would suggest with Linux , but I would clean install /DB your OS ; just to see how it runs , you don't even need to re-install your apps , just pull them from the source directory .

" If I was your tech guy and mentioned Virus my defence would be straight up :D "

but if the error started after (x) amount off time , I would not rule out the possibilty of a corrupt systems dir and generally a ram error is indicated at boot-up

JBT27
04-24-2010, 02:24 PM
Nightmare for sure. XP64 ..... right or wrong, it's worked until recently, and pretty well.

Are you saying that if it is a RAM error, I'd be seeing it at boot-up? I reckon I'd be seeing it more than this anyway.

Re-installing is starting to look like a possibility for sure.

Thanks.

Julian.

OnlineRender
04-25-2010, 04:48 AM
usually a RAM error / failure / incompatiple RAM give's an indication at the start of boot-up , Ha originally I had written down XP as your main OS but i thought you would be running higher than 4GB ?

if it's XP ,what SP ? , if you get a clean install and everything goes back to normal , its a windows error , but I thought about this last night , it could be video card failure ... dont suppose your or ATI readon series ?? it could be the GFX card locking up ! lol working out the problems 1 step @ a time :D goodluck

JBT27
04-25-2010, 04:59 AM
I think I'm erring away from the RAM error thing ..... not ruling out anything, but it's not looking alot like that.

Yes, XP64 SP2 - the machines are about a year old and the choice to go with XP64 was pretty much based on comments on these forums. But, they are 8Gb machines, and once this deadline is passed, maybe we'll rethink the OS .....

Video card ..... yeah ..... well, we tried putting the self-same card from the other (identical) machine into the dodgy one, but the problem persisted. Right now, however, I am wondering if the driver is getting messed up for some reason - it's been OK, version 182.50 from March 2009 by the looks of it, but the latest is 197.45.

The thing that occurred to me was that although the entire screen appeared to be locked up, maybe it's the refresh just not happening, or sticking ..... like I say, it never crashes, just freezes for awhile.

Really don't need this on top of a deadline ..... but that's when it always happens ...... not like I'm not used to it :D

I appreciate the ideas - yep, one by one :D

Julian.

JBT27
04-25-2010, 05:19 AM
Well, updated the video card driver ...... not that it's made any difference :grumpy:

Julian.

JBT27
04-25-2010, 05:33 AM
This is probably very dumb, but two or three weeks back, I installed the latest CORE and updated the Sentinel drivers, because you are required to ..... no chance of any of that affecting this is there?

Julian.

OnlineRender
04-25-2010, 05:37 AM
http://puppylinux.org/ :D

danielkaiser
04-25-2010, 11:08 AM
Does your event viewer show any errors?

OnlineRender
04-25-2010, 11:53 AM
This is probably very dumb, but two or three weeks back, I installed the latest CORE and updated the Sentinel drivers, because you are required to ..... no chance of any of that affecting this is there?

Julian.


hmm doubt it !

JBT27
04-25-2010, 11:55 AM
Does your event viewer show any errors?

No, I wouldn't say so - I have one Application error reported for today, which was a Layout crash, and the only System errors are generated by W32Time, which is not relevant to this, afaik. There are no Security errors.

Hmm ..... I've gone way past baffled on this one, but it's bloody frustrating and a bit nightmarish.

Wondering about video card driver settings, which I haven't adjusted at all, but even so .....

But it is system wide as not just AE RAM previewing will cause a lock-up, but a LW F9 render, or any of the other stuff I mentioned above.

Looks kind of like a bottleneck somewhere .....

Maybe it's a network thing ..... ??

Julian.

JBT27
04-25-2010, 11:56 AM
hmm doubt it !

Thank god! Everything was a suspect ..... at least I can let that one go ..... probably :D

Julian.

OnlineRender
04-25-2010, 12:08 PM
it could be your netgear drivers ! ive seen people have problems with freezing and the netgear was the problem

Dexter2999
04-25-2010, 12:19 PM
try a freeware RAM dumper. If you are experiencing hangs after things that are using RAM, it could be your system is slow to dump or refresh the RAM. Kind of like a temporary Memory leak. These little programs take up like 2Meg.

Good luck.

JBT27
04-25-2010, 12:47 PM
try a freeware RAM dumper. If you are experiencing hangs after things that are using RAM, it could be your system is slow to dump or refresh the RAM. Kind of like a temporary Memory leak. These little programs take up like 2Meg.

Good luck.

Thanks - I'll take a look at those.

Julian.

Martin Adams
04-26-2010, 05:25 AM
Things that I've seen in the past that have cause some form of hangs are:
- Real time virus scanner, try disabling it completely if you have one.
- Virus scanner (AVG in particular) corrupting during an update and brings the whole system to it's knees.
- Hard drive issues which causes freezes when it's in use (maybe the larger use of RAM is paging something out?)
- I have also had power issues (i.e. me not plugging in an extra power cable to the CPU directly - but that caused an actual crash when under high usage)
- Might want to see if you have any particular power plans that limit the CPU or try to change something after a period of time (along with screen saver)

Hope you get it figured out soon!

JBT27
04-26-2010, 06:50 AM
Things that I've seen in the past that have cause some form of hangs are:
- Real time virus scanner, try disabling it completely if you have one.
- Virus scanner (AVG in particular) corrupting during an update and brings the whole system to it's knees.
- Hard drive issues which causes freezes when it's in use (maybe the larger use of RAM is paging something out?)
- I have also had power issues (i.e. me not plugging in an extra power cable to the CPU directly - but that caused an actual crash when under high usage)
- Might want to see if you have any particular power plans that limit the CPU or try to change something after a period of time (along with screen saver)

Hope you get it figured out soon!

Thanks. The tough thing about this week and weekend is the deadline, so I'm having to resolve to get through with the problem - I'm going to use AE6.5 that I still have on my dual-core, which should do fine for this job, and put up with the lesser lockups that LW causes when rendering with F9/F10.

Last night, I let AE CS4 run a RAM preview for 5 mins, just playing it. At the end, when I tried to do something in AE, the whole AE interface just blanked out to grey and I had to force quit it. As other things seem to cause these lockups besides AE, I can't entirely look to that for the root of the problem, but it is the heaviest user of RAM for sure.

No AV of any kind on the box - nor screensavers or anything like that - it's supposed to be a lean and mean production machine :D ..... so much for that theory :)

Thanks for the additional ideas - I really am at a loss to where to start troubleshooting next. Maybe I will go for a reinstall of the OS and see if that is the culprit. As it's system wide, it's clearly OS or hardware, rather than just AE.

Julian.

OnlineRender
04-26-2010, 07:00 AM
reinstall the OS but keep the old OS *Dual Boot* that way you can still run AE or other apps direct from the original source directory .

or DOS c:/format /s :D kiddin dont do that , I got banned from Comet for LIFE for doing that :D

JBT27
04-26-2010, 07:54 AM
Haha :D ..... I wouldn't do that anyway! Banned for life from Comet is impressive though.

Also your idea about Netgear drivers ..... that machine is only using onboard gigabit though and the OS drivers, so again it could be that but locked in with the OS as being the issue ..... if it is.

I can see that reinstall coming on next week.

Julian.

OnlineRender
04-26-2010, 08:44 AM
Ye didn't go down to well :D that's what they get for leaving there syst3ms open :D , it was about 6 years ago , when packerdbells where the in thing :D and the people were real salemen :D

JBT27
05-28-2010, 01:58 PM
Just bumping this because I've still got the problems, but having finished that job that meant I couldn't mess about with things, I've stumbled on something but need some help please figuring it through.

So the general consensus was that this looked like a RAM issue, and I said that I'd run Memtest and got errors within seconds of starting the test.

So today I ran Memtest again, for several hours, on our four quad-cores and our one dual-core.

The two basic quad-cores for rendering and the dual core ran Memtest without reporting errors, and the Memtest ran for something like 350% - 400%.

On my quad-core that I was talking about, I ran it for half an hour at first but reached something like 70 errors between the two Memtest windows I had open, and that was only around 23% of the test. So I shut down the machine, took two strips out and tested again - errors. So, took those out and put the other two back in - errors again.

Thing is, we've got two of these machines, identical, and while I was doing these tests on mine, I ran Memtest on the other. That was generating alot of errors as well.

We've had these two machines for a year, and to be honest they've been mostly OK. The temp freezing/locking-up problem with mine is since March or so, but to be honest I never tested the RAM until then, and only today on the other one. We had the machines built by a local shop who builds alot of stuff for us, and assumed when we got them that all would be OK. We've done several jobs without major problems, so no real reason to worry.

Except now, bottom line, we have two identical machines, same mobo, same RAM, both throwing out tens of errors over a test lasting three hours or so, without even reaching a 100% of the test cycle. Our two lesser quad-cores on the other hand ran up to 400% of the Memtest cycle without errors, likewise the dual-core.

We've got 4x2Gb modules in each box - OCZ PC2 8500-Platinum Edition-
PN: OCZ2P10664GK-5-5-5 2.2V-2Gb.

The mobo is Gigabyte GA-EP45-UD3.

In BIOS, the RAM settings are Auto - the board detects what the RAM is and sets the appropriate settings.

So what do you reckon the most likely thing is? Just plain subtley incompatible? A complete bad batch of RAM, but not so bad that we get errors that are major killers?

Not sure what to do here - £400 of RAM a year old and the computer building guy saying it's still worth checking the OS and the drive. With this much error on two identical machines, surely that's not coincidence and surely it has to be something between that RAM and the mobo?

I'd appreciate any more help and ideas because we do need to get this sorted.

Thanks.

Julian.

monovich
05-28-2010, 10:39 PM
I had random problems not unsimilar to this and narrowed it down to bad ram. I had my machine overclocked and I'm wondering if I overheated it over time or something. The ram was probably 1.5 years old. Only one stick went bad and taking them all out but one stick fixed it, then I added another until the crashing came back.

maybe try loading default values to your bios if you've done anything custom to it.
are you overclocked?

if you run the memtest with only one stick, do you still get errors?

Danner
05-28-2010, 11:01 PM
I have similar issues on a double quad machine I use at work. The thing is it doesn't happen untill I have a ton of stuff going on or the machine has been rendering while I work on a separate modeler/photoshop etc session. After a reboot it works fine for a day or two then it starts to hang again.

Have you eliminated heat as a cause? Ram gets hot too, some of the high end ram comes with heatsinks and fans.

Oedo 808
05-29-2010, 12:59 AM
I would always set the RAM manually to match it's voltage and timings as advertised, even this may have to be adjusted (by slackening the timings) if you are overclocking.

For anyone who has a Gigabyte board you can try asking at the TweakTown (http://forums.tweaktown.com/) forums, they have an official Gigabyte tech support (http://forums.tweaktown.com/f69/) section.

I don't know if they still do it, but when I built my Q6600 system a while back I got a full run down of what mobo settings to try for a stable overclock, it was very, very helpful, Lsdmeasap is a star.

No two sets of components are the same, what works for a stable OC on one system may not work for an identical set-up right beside it, so getting a system stable can involve a fair bit of trial and error, there are no golden values that will guarantee success.

When I OC'd, my initial overclock appeared stable, it passed 12hr+ of Prime95 and memtest, and the only problem I ever noticed was that FPrime would sometimes stop updating when I moved the camera, so I tried Intel Burn Test and it failed my OC in under a minute or two, after refining a few setting it worked fine. It's was confusing at first, but memtest will fail perfectly good memory if the auto-detect isn't giving it enough juice (I think they would rather it under-volt than fry cheaper RAM modules), or if it's running under an unstable overclock it will also show errors with good RAM.

RLM
05-29-2010, 01:16 AM
To the OP I think the problems you have could possibly be resolved in the BIOS settings.

You're using a Q9550 so I'm presuming the motherboard is a 775. It's notoriously difficult with these boards to get 4 sticks of ram stable.

For one you may need to increase the Vdimm a tad over the recommended spec voltage. i.e. if the spec is 1.8v then try 1.85v

Secondly look at the timings. You may need to loosen them up a bit. So for instance if the memory CAS RAS timings are set at something like 7-7-7-25 try changing them to 8-8-8-25

In addition if the command rate is set to 1T(sometimes 1N) try loosening it to 2T(2N).

You could also possibly do with a marginal bump in your NorthBridge voltage as well. So if it's currently 1.3v try 1.35v

That said it would be a good idea to check the brand of your memory and motherboard you have and visit the appropriate forums to see what's recommended there.

Just one last thing it would be a good idea to download the following

Prime95(x64)
http://files.extremeoverclocking.com/file.php?f=205

You want for starters to run the blend torture test and really it needs to run for a few hours or so without errors.

RLM

ps. Out of interest what model is your motherboard and what type of memory do you have?

RLM
05-29-2010, 01:20 AM
PS. You need to get this fixed v.soon. With unstable memory at the v.least you're going to end up with a corrupt OS.

JBT27
05-29-2010, 03:48 AM
Thanks everyone - and for the timings detail RLM - appreciated.

So I'll admit we paid this guy to build these machines, but with hindsight stupidly believed they would be ready to go and stable.

Looking at the BIOS now, our OCZ RAM modules have a spec that includes 2.1-2.2V, whereas the DRAM Voltage in BIOS was set to 1.8V - as he said to make sure they don't burn out ..... that's kind of looking like a dim thing to have told us now - no, we are not big on hardware, so you'll have to forgive the naivety :)

These are our RAM modules - 4 x 2Gb per machine:

http://www.ocztechnology.com/products/memory/ocz_ddr2_pc2_8500_platinum_4gb_edition_dual_channe l

So far, I have adjusted the DRAM Voltage to 2.2V and CAS RAS timings to 5-5-5-18 (it was auto set to 5-5-5-15).

The Command Rate is still set to 0 (I can set 0-3).

Memtest is still giving errors though.

Yes, a 775 motherboard:

Gigabyte GA-EP45-UD3

It wasn't our intent to overclock. I suspect Oedo 808 may be right in suggesting that Memtest is flagging errors through poor tuning as opposed to bad RAM - if it's bad RAM, then we bought 8 modules that are all bad at the same time. Wouldn't rule it out, but .....

So, if these boards are difficult to get stable with 4 modules, I guess I have alot of tweaking ahead. I'll hunt around and see if I can find recommended timings for this kit, but any more thoughts and pointers will be very welcome.

Thanks.

Julian.

Oedo 808
05-29-2010, 04:09 AM
RLM is right, I'm sure running four modules means you'll need to bump the northbridge voltage, but I don't want to guess at settings based off a different board, my advice is to fill out a BIOS template here (http://forums.tweaktown.com/f69/gigabyte-bios-templates-give-us-your-settings-33613/), and post it along with your problem at the aforementioned link here (http://forums.tweaktown.com/f69/). Good luck.

JBT27
05-29-2010, 04:27 AM
OK thanks - I'm doing that now.

Just ran Prime95 on both machines, which did 2 to 3 tests in 3 to 4 minutes each, but with each of the four 'workers' in the test, on both machines, an error was found and testing stopped.

So, I'll see what comes of posting on TweakTown.

Julian.

RLM
05-29-2010, 04:47 AM
Right for starters here is a template copied from here

http://www.clunk.org.uk/forums/reviews/14031-gigabyte-ga-ep45-ud3r-living-review-2.html

Might be worth a good read through that thread as there maybe poeple experiencing similar issues.



GIGABYTE GA-EP45-UD3R
[u]MB Intelligent Tweaker(M.I.T.)
Robust Graphics Booster ...............:
CPU Clock Ratio ..........................:
Fine CPU Clock Ratio.....................:.
CPU Frequency ...........................: Mhz
Clock Chip Control
Standard Clock Control
CPU Host Clock Control..................: [Enabled] [Disabled]
CPU Host Frequency (Mhz) ............:Mhz
PCI Express Frequency (Mhz) .........:Mhz
C.I.A. 2......................................: Disabled
Advanced Clock Control [Press Enter]
CPU Clock Drive...........................: mv
PCI Express Clock Drive.................: mv
CPU Clock Skew (ps)....................: ps
MCH Clock Skew (ps)...................: ps
DRAM Performance Control
Performance Enhance...................: [Turbo]
Extreme Memory Profile (X.M.P.).....:
(G)MCH Frequency Latch...............: [Auto]
System Memory Multiplier ..............:
Memory Frequency (Mhz) ..............: Mhz
DRAM Timing Selectable ................: [Manual]
Standard Timing Control
CAS Latency Time..............................:
tRCD .........................................:
tRP...........................................:
tRAS..........................................:
Advanced Timing Control [Press Enter]
tRRD...........................................:
tWTR..........................................:
tWR............................................:
tRFC...........................................:
tRTP...........................................:
Command Rate (CMD) ....................:

Channel A

Channel A Timings Settings [Press Enter]
Static tRead Value.........................:Auto
tRD Phase0 Adjustmen.......................:Auto
tRD Phase1 Adjustment......................:Auto
tRD Phase2 Adjustment .....................:Auto
tRD Phase3 Adjustment......................:Auto
Trd2rd(Different Rank).....................:Auto
Twr2wr(Different Rank).....................:Auto
Twr2rd(Different Rank).....................:Auto
Trd2wr(Same/Diff Rank).....................:Auto
Dimm1 Clock Skew Control...................:Auto ps
Dimm2 Clock Skew Control...................:Auto ps
DDR Write Training ...........................: Enabled
Channel A Driving Settings [Press Enter]
Driving Strength Profile...................:[AUTO],[667Mhz],[800Mhz],[1066Mhz],[OC-1200],[OC-1333]

Data Driving Pull-Up Level.................:Auto
Cmd Driving Pull-Up Level..................:Auto
Ctrl Driving Pull-Up Level.................:Auto
Clk Driving Pull-Up Level..................:Auto

Data Driving Pull-Down Level...............:Auto
Cmd Driving Pull-Down Level................:Auto
Ctrl Driving Pull-Down Level...............:Auto
Clk Driving Pull-Down Level................:Auto
Channel B

Channel B Timings Settings [Press Enter]
Static tRead Value.........................:Auto
tRD Phase0 Adjustmen.......................:Auto
tRD Phase1 Adjustment......................:Auto
tRD Phase2 Adjustment .....................:Auto
tRD Phase3 Adjustment......................:Auto
Trd2rd(Different Rank).....................:Auto
Twr2wr(Different Rank).....................:Auto
Twr2rd(Different Rank).....................:Auto
Trd2wr(Same/Diff Rank).....................:Auto
Dimm1 Clock Skew Control...................:Auto ps
Dimm2 Clock Skew Control...................:Auto ps
DDR Write Training .........................: Enabled
Channel B Driving Settings [Press Enter]
Driving Strength Profile...................:[AUTO],[667Mhz],[800Mhz],[1066Mhz],[OC-1200],[OC-1333]

Data Driving Pull-Up Level.................:Auto
Cmd Driving Pull-Up Level..................:Auto
Ctrl Driving Pull-Up Level.................:Auto
Clk Driving Pull-Up Level..................:Auto

Data Driving Pull-Down Level...............:Auto
Cmd Driving Pull-Down Level................:Auto
Ctrl Driving Pull-Down Level...............:Auto
Clk Driving Pull-Down Level................:Auto


{U]Motherboard Voltage Control{/U]
Voltage Type.………... Normal
CPU
Load Line Calibration......................: [Enabled] [Disabled]
CPU Vcore..................................:v
CPU Termination…..... 1.200V*..............:v
CPU PLL...............1.500V*..............: v
CPU Reference.........0.755V*..............: v
MCH/ICH
MCH Core..............1.100V...............: v
MCH Reference.........0.800V...............: v
MCH/DRAM Ref..........0.900V...............: v
ICH I/O...............1.550V...............: v
ICH Core............……1.100V...............: v
DRAM
DRAM Voltage..........1.800V...............: v
DRAM Termination......0.900V...............: v
Channel A Reference...0.900V...............: v
Channel B Reference...0.900V...............: v
{U]Advanced Settings[b]{U]
Limit CPUID Max. to 3......................: [Disabled]
No-Execute Memory Protect..................: [Disabled]
CPU Enhanced Halt (C1E)....................: [Disabled]
C2/C2E State Support.......................: [Disabled]
x C4/C4E State Support.....................: [Disabled]
CPU Thermal Monitor 2(TM2) ................: [Enabled]
CPU EIST Function..........................: [Disabled]
Virtualization Technology..................: [Disabled]
[b]Note: * refers to 45nm CPU


OCZ forum for gigabyte motherboards
http://www.ocztechnologyforum.com/forum/forumdisplay.php?223-Gigabyte-Motherboard-Section&order=desc

For starters I would try the sticks at 6-6-6-20 timings. When using 4 sticks you do have to make some sacrifices, but I doubt you will notice a big difference in day to day usage.

RLM

JBT27
05-30-2010, 02:51 AM
Well, done alot of checking and tweaking and more RAM testing over the past day or so - no change to the reported problem.

The thread over on TweakTown is here:

http://forums.tweaktown.com/f69/ga-ep45-ud3-4-x-ocz-pc2-8500-unstable-ram-39816/#post357809

Just finished trying some of what kick suggests, but no visible change - even managed to have the machine refuse to boot.

So at this point, well, I really don't know.

As to the locking up problem - a friend actually suggested that these machines with 500W PSUs are substantially underpowered. They made the point that if my PSU is failing, or too low, the GPU will have slowed to match. Which makes me wonder if the lock-ups I'm seeing, effectively screen freezes where I cannot click anything for alot of seconds, is the card's buffer trying to clear after rendering, either in LW or AE.

Either way, the RAM errors thing with Memtest and Prime95 might just be spurious, and actually there's nothing wrong with the RAM, ie. the problem is elsewhere, but feeding down to the RAM testing.

Reading through what kick says, it kind of has the look that the machines are badly configured and maybe even shouldn't work that well at all - which is a bit of a problem as we've managed alot of intensive projects with them through the past year. As I just found, moving them away from what they were set to actually caused one not to boot.

So we've got our computer shop thinking this is more OS related, and now the issue of the PSU ..... :(

Julian.

RLM
05-30-2010, 03:10 PM
Right need to take it back to one stick and at least see if you can get that stable.

Did you try 6-6-6 timings with a command rate of 2t?

Regards the PSU, I agree. It's not man enough for your setup and could in fact end up killing your hardware.

A really decent site for PSU reviews
http://www.jonnyguru.com/

After using both Enermax and a silverstone OP1000 I can recommend the thermaltake toughpower PSU's. Modular and with decent components.

The 1KW model I've now had for about 2 years. I've had it powering a qx9650 to 5ghz(benching) and 4.5ghz 24/7, 4870x2 heavily overclocked in crossfire and watercooling and it's held up well. It really is the most important part of your kit.

I don't really know about your board. You do find that certain boards favour certain sticks of ram. I think I'm right in saying that some modules come with micron chips and other come with samsung (epida?). Certain boards favour one or the other.

Regards the corrupt OS that would be a symptom of using an unstable setup. As I said at the very least it can lead to a corrupt OS, worst it could lead to corrupting your hard drives and partitions.

edit: Will check out your thread on tweaktown. Did you post on the OCZ forum?

RLM

RLM
05-30-2010, 03:15 PM
Just spotted something

DRAM Performance Control
Performance Enhance ------------------- : [Turbo]

What are the other settings for performance enhance?

Try taking it off turbo and putting it on something less strenuous

RLM

RLM
05-30-2010, 03:24 PM
There's other things as well. For instance on your q9550 you're using a multiplier of 8 (actually it's 8.5x333). You might find that it actually prefers 9 (9x315).

I also see you're still favouring a cas of 5

CAS Latency Time........................ 5

This Cas setting is a biggy. Rather than loosening up the other timings loosen this one to 6.

Oh and TRFC as well. You could try loosening that to 92(I think) or 100.

EDIT 2) And another

Static tRead Value.........................: 6 Auto
This one I remember being pretty significant, and at the moment that looks pretty tight to me. Try loosening that one to 8 or 9 even.


I mentioned this before, I really don't think you will notice a difference in performance by looseing up your ram timings. When it comes to rendering times the difference is negligible (CPU is king). You really only tend to notice a bit of a difference when it comes to FPS in benchmarks like 3dmark.

Last edit(I think). As suggested in the forum MCH may need to be 1.2-1.24. I'd certainly try upping the MCH a tad. You did mention freezing but this maybe a combination of factors (like a weak PSU for instance).

RLM

JBT27
05-31-2010, 02:16 AM
RLM - thanks very much for all the info and ideas.

As to testing individual sticks, I've been doing that - tested them in pairs, on both machines, but as I mentioned over on TweakTown it was quickly clear that it wouldn't just be one stick at fault, which in turn makes me think this is either a massive incompatibility, or BIOS settings.

Before saying more about the RAM, the original post in this thread, about the odd lock-ups after RAM intensive tasks, like rendering, or AE RAM previewing, etc., I seem to have reduced or even fixed by deleting the Paging File off the boot drive. Now before anyone jumps on my head :D for having one on there anyway, so far over many years doing work like this, including that big HD broadcast job late last year, that wasn't an issue, even given having a single drive in that machine. Now there are two drives, I've put a Page File on the other drive and set the boot to have none.

That said, in hunting for solutions, clearly I've flagged up this RAM issue, and in turn the potentially underpowered system. We'll sort bigger and better PSUs this week.

But so far, even with the suggestions I have tried so far, I cannot find settings that stabilise the RAM.

I did try 6-6-6 and 2t, but still got errors. Switching from Turbo to Standard on the Performace Enhance likewise.

Increasing MCH didn't work. Increasing DRAM voltage didn't work. Increasing both caused the machine not to boot.

On the Multiplier, my top setting is 8, and the only fine tuning options are 0 or +0.5, so I cannot hit 9 anyway.

We recently bought a hardware media player, a ViewStream, for a client. Although the manual and website state that any compact flash card should work, it did clearly say that it prefers Sandisk. Because we were in a hurry and a card doesn't come with the device, we bought the only card that Maplin down the road could offer - not Sandisk. It didn't work, or rather it kind of did, then got flaky, then stopped altogether. When we did get Sandisk, it just worked.

So I have to wonder if this is the same situation. As I say our other two un-tuned quads and the old dual core run Memtest till the proverbial cows come home, without errors. So no matter what I do in BIOS, they just ain't going to work.

Seeing as we spend so much money with that shop, I was hoping he might let me try a couple of sticks of other RAM, but it doesn't look like that's going to happen, so I'd need to drop just under £100 to try a couple of the Transcend sticks he has in right now.

The other thing is that these machines have been in heavy use for a year, but the problems you might expect with bad RAM and failing OSs and the rest, haven't really shown. That's not to be complacent, but I'm just saying that but for this locking-up thing (which seems now to be Page File related), nothing odd would have got flagged up.

I'm going to post in the OCZ forum shortly - there's a chance someone on there has the same combo, with a bit of luck.

I don't want to hit a point where the machine won't boot - or is that not an issue? Like yesterday, when what looked like a minor and isolated change to BIOS stopped it booting.

OK, so I'll post to the OCZ forum now.

Julian.

RLM
05-31-2010, 02:47 AM
From the clunk thread. I wouldn't right the sticks off quite yet.


I have the UD3P (very similar, except PCI/PCIE slots) running fine with the OCZ 1066 CL5 4GB Platinum kit, which is basically the same as the reapers but uses normal heatspreaders, so you should't have any issues with the reapers. Be aware though that the reapers may block you from using certian large tower CPU heatsinks, which is what you want if you're not going for watercooling.

Do try the static tRead as well it is a significant setting.

Edit (It's been a while since the 775s for me. If you look at your Memset 4.1 screengrab. Perfomance Level - 6. That's your tRead)

Edit2 (I've had a look at a few bios configs so far and haven't seen anyone going below pl7. Even that seems to be pretty tight. Definitely try 8 or 9

Comments from supershanks on the clunk forum.


Yep the Static tRead Value or Trd value is ny far the biggest stumbling block in getting this mobo clocking well. Think we've all been victims of it at 1 time or another)

This was the thread I was looking for all 200+pages dedicated to the Gigabyte UD3 P45 Series.

http://www.xtremesystems.org/forums/showthread.php?t=205132


We'll sort bigger and better PSUs this week.

That's good news:)

If you check out jonnyguru, you'll see he does some very thorough testing so well worth a look.

RLM

RLM
05-31-2010, 03:10 AM
I'm just scanning through the xtreme thread. The first one that stands out

http://www.xtremesystems.org/forums/showpost.php?p=4103677&postcount=5092


here is my 24/7, vcore 1.33, vtt 1.36, mch 1.38, vdim 2.16, pl9, 2.00d strap

pl9 is performance level (static tRead).

vtt is CPU Termination. 1.3v should do, but keep it under 1.4v

Another couple of posters pointed out that when these settings were left to AUTO, they had problems so opted to set them to NORMAL.

DRAM Termination .…0.900V............
Channel A Reference 0.900V............
Channel B Reference 0.900V............

More reading....

RLM

JBT27
05-31-2010, 03:16 AM
Thank you - checking through that as well - just finishing post to OCZ.

You see, what's making me nervous is the combo of settings - do you try one and not another, as it were, or one AND another then another ..... geez - there must be thousands of combinations at this rate :D

So when you say about trying the static tRead, do you mean with the other suggestions, or just on its own?!

Sorry if I'm being dim - did say I'm pretty new to this, and all I want is my boxes working and stable and not slowly committing suicide :D

Julian.

JBT27
05-31-2010, 03:50 AM
Here's the OCZ thread:

http://www.ocztechnologyforum.com/forum/showthread.php?73658-GA-EP45-UD3-2-x-OCZ-DDR2-PC2-8500-Platinum-Edition-4GB%282x2048%29-2-Identical-Machines-Multiple-Errors&p=521307#post521307

Julian.

RLM
05-31-2010, 03:57 AM
You see, what's making me nervous is the combo of settings - do you try one and not another, as it were, or one AND another then another ..... geez - there must be thousands of combinations at this rate

It's fun'n'games for sure.

My thinking at this point.

Sensibly you want to hold on really until you have the new PSU in there. You can then at least right that off out of the equation.

In the meantime have a look through the xtremesystems thread and see if you can find similar builds to your own. Quite often people post their bios settings and this would give you a good base to work from.

Now in that thread most are overclocking. What I would do is see if I can find a post with maybe a modest overclock on a q9550 say 3.4-3.6. Something that's posted as a 24/7 on air.

Without altering your FSB of 333 x 8.5 (i.e. overclocking) look at the other readings such as voltages and use them as a base. Like these settings for instance.

Load-Line Calibration.................: [disabled] (this controls vdroop)
CPU Vcore..............1.23750V........: [???]
CPU Termination........1.200V.........: [???]
CPU PLL................1.500V............: [???]
CPU Reference..........0.760V.........: [???]
>>> MCH/ICH
MCH Core...............1.100V............: [???]
MCH Reference..........0.760V.........: [???]
MCH/DRAM Reference...0.900V.....: [norm]
ICH I/O................1.500V...............: [norm]
ICH Core...............1.100V.............: [norm]
>>> DRAM
DRAM Voltage...........1.800V...........: [2.20v]
DRAM Termination.......0.900V.........: [norm]
Channel A Reference....0.900V.........: [norm]
Channel B Reference....0.900V.........: [norm]

Regards Ram. You certainly want to start off with loosening up your timings 6-6-6-20 TRFC 90ish, static tRead 9 and turbo off. You could even try 7-7-7-20. You can always tighten these timings back up again once you attain some stability.

It will take a bit of time to get it stable. Maybe a couple of days, but once that is done you then forget about it.

RLM

JBT27
05-31-2010, 05:57 AM
OK - great, thank you.

Interesting definition of 'fun' :D

So I just tried the 6-6-6-20, still got errors, so changed to 7-7-7-20 ..... the machine struggled to boot; there's no beep or anything on this box, but there is a long Cylon-type light on the front as it powers-up. So after changing to 7-7-7-20, that light came on, waited, then clicked off, then on again, then off, then on, then off, then on, then it restarted ..... kind of unnerving :D

So, as per your very sensible idea - PSU first, then take it from there.

One thing: running at 6-6-6-20, it took alot longer for the errors to start appearing .....

Case in point - I've set defaults back to Optimised, running MemtestPro, and I have four instances running. One has only hit 20% coverage, but with 3972 errors, another at 19.4% with 3886 errors, the third at 278.5% with 3366 errors, with the fourth at 20.5% with 4224 errors.

It's worrying that these errors are not manifesting in everyday use of the machines, but other than the odd crash, which may or not be RAM, and given the Paging File issue, there's not much else.

I note also here:

http://www.giga-byte.com.tw/products/product-page.aspx?pid=2920&dl=1#driver

that the BIOS is up to F9, versus our F3, and the latest specifically cites improved memory compatibility .....

But I'll go back to 6-6-6-20, with some of the other settings from before, now and see how Memtest actually does this time.

Julian.

RLM
05-31-2010, 07:24 AM
It's worrying that these errors are not manifesting in everyday use of the machines, but other than the odd crash, which may or not be RAM, and given the Paging File issue, there's not much else.

It could still be corrupting the OS and it could be just a matter of time.

I've just been reading a bit of supershanks first posts on the clunk forum, and it's pretty informative as far as suggestions go.

Just to confuse matters for you, you have various straps. I think for your current setup of 333fsb with 1066 memory you need to set the memory multiplier(strap) to 2.5b(This is unique to the gigabyte boards). You'll see supershanks has posted a chart on the first page.

Here's another thread by a guy called George on i4memory. He really does know his onions. I've been a member of that forum for some time.

http://i4memory.com/f90/gigabyte-ga-ep45-extreme-rev1-0-review-8339/

He's running an e8600 on that board, so it's not all applicable, but still worth a read.

For instance on the 5:6 divider (2.4b) I think you're using

http://i4memory.com/f90/gigabyte-ga-ep45-extreme-rev1-0-review-8339/#post93992

I see his first comments.

First up is 2.40B divider (5:6) and it seems by default the Static tRead value that is set by default for 2.4B divider is too tight. At 9.5x333FSB 2.40B 400Mhz 5-5-5-15 at 2.00v vdimm with CPU VTT and MCH set at 1.20v and 1.30v respectively, Static tRead value is automatically set to 7 according to what the bios says and results in a power down/reboot and reset cycle when you save and exit bios. Probably, due to the low MCH voltage somewhat. In his first test he's actually set tRead to 10.

You are setting the static tRead for both channels A and B?

I remember quite a few years ago when I bough a DFI Expert and some corsair LL pro memory. Had real troubles getting it to work. Even had the board replaced with a new one. Had the same troubles with the new board, I then experimented with one particular memory timing, loosening it up a bit and it was absolutely fine after that.

Regards the bios. Leave it for the moment. Latest isn't always the greatest. You'll probably find a particular favored bios in the xtreme posts. You want your new PSU in first and at least the kit stable with one stick of ram. Trying to flash the bios while the system is unstable is a bad idea. You could end up corrupting the bios, and in which case the board would have to be sent back to rectify it. That or hot swapping, but I won't go there:D.

RLM

JBT27
05-31-2010, 07:30 AM
At a little over 20% coverage, three MemtestPro instances are reporting 1212, 1008 and 1242 errors, the fourth instance at 283% is 1354 errors. So down on last time with the Optimised Defaults.

These lesser errors are with 6-6-6-20, tRFC 90, Static tRead 9.

Julian.

JBT27
05-31-2010, 07:46 AM
Agreed on the potential for slow but sure corruption. Though without wishing to be pedantic, I am wondering what the definition here of 'RAM errors' actually is, because clearly our systems are not at all stable in that sense, but on the other hand over a year we have not seen major symptoms of duff RAM .....

System Memory Multiplier is set to Auto - it's not reporting exactly what. There's a 2.50A but not 2.50B.

Yes, setting tRead for Channels A and B.

I just upped the DRAM Voltage from 1.80 to 1.82 - errors came thick and fast - so dropped it back to 1.8. Though this for sticks rated for 2.1-2.2 ..... ??

No, I was not planning on flashing the BIOS just yet, if at all :)

The whole point is to stabilise and keep us working, not sending boards back with red faces because we screwed up :D

Thanks for the extra threads - I'll go and read those now.

Julian.

Oedo 808
05-31-2010, 09:53 AM
Going through this process drove me nuts enough when I was overclocking, I can't imagine how frustrated I'd be just trying to get it stable at stock.

It's not exactly comparable because I'm oc'ing, but if I set the System Memory Multiplier to auto it would clock the memory at 1275 instead of 1020, that would cause my system to be unstable and no mistake :p though admittedly auto should do better for stock settings.

Having said that, is the MCH frequency latch set manually to something? Because that setting is linked to the memory multi and your choices there will be limited according to the latch setting you have, that is one I do have on auto. Only having 2.5a doesn't sound right, though I can't remember what other factors would affect this.

If you do decide to update the BIOS (I did for my board), I would put the BIOS on a USB pen and update it from that within the BIOS itself, I don't think it need be a bootable USB but I'm afraid I just can't remember, and yes, updating your BIOS is always 'cross your fingers' moment even though I've yet to have a bad flash. I would not use a Windows based utility, especially not if I was already having issues.

Did you set your PCI Express freq to 100? You don't want this being boosted.

I will check mine and be more specific if you need, but in the advanced memory settings there should be a section that refers to Driving, in that will be a Driving Strength Profile or something similar, you might try changing this to 1066 (for both A and B). And for what it's worth my Static tRead is 8.

I may be wrong but I can't see the RAM not needing more juice eventually, I would suspect that the instability is coming because incorrect settings mean it is already working too hard and that's why the lower volts relieve the symptoms, but I'm no expert I just OC'd my system, nothing more.

It has been a year and a half or more since I last touched this stuff, so I'd filter anything I say through what RLM is advising (then you can blame him for the smoke :p).

As has been mentioned, upgrade your PSU first (if you didn't already, I'm far too lazy to read it all), a system at stock isn't going to require all that much power really, but if the PSU quality is mediocre that can affect everything. After all this, if no luck, I would think about updating that BIOS, things like these are part of why they get updated.

Hmm, I think this thread has just increased the life expectancy of my current system. :)

JBT27
05-31-2010, 12:07 PM
Well, I'm getting a bit of a taste for this stuff with all this, and all the help I'm getting, which is greatly appreciated ..... but that said, I'm kinda pinching myself thinking 'how come I'm having to get my head around all this on machines that are not overclocked and were not required to be, when all I want to do is get on with my pictures?!' ..... frustrating doesn't begin to describe it :D

However, happy to be getting to grips with the beginnings of such things - all new to me, this stuff.

MCH Freq Latch is not set manually - don't know what it is right now, but it is on Auto, afaik.

PCI Express Freq definitely set to 100, manually.

I have to say, I have a dread that we're going to have flash the BIOS in the end - bigger and better PSUs aren't going to go amiss anyway, but I'm wondering how is that the process of flashing the BIOS is such a finger-crosser?!! Seriously, you'd have thought 'they'd' make it easier and reversible by now. But there you go, what do I know? :)

This thread might have increased the life expectancy of your machine ..... I suspect it's actually shortened my life expectancy! :D

Julian.

JBT27
05-31-2010, 12:09 PM
Got the first response from an OCZ tech over on that thread:

'Are you on the most recent BIOS already? If not flash to that.

Memory timings should be 5-5-5-18(CAS-TRCD-TRP-TRAS), tRFC 60 with the memory voltage at 2.1V and perhaps a bump in the chipset voltage as well.'

Hmm - well, I'll give it a try. Pending new PSU notwithstanding, the default settings aren't that anyway, so I guess I should try it.

Julian.

JBT27
05-31-2010, 12:29 PM
Of course these are the official spec of the RAM - still getting alot of errors with those settings, which are tighter than the ones earlier.

So I think I'll wait for those PSUs now, as RLM suggested. I've had a slight improvement in the error count today, but it's far from perfect, so I reckon it's time to sort other bits out then start again.

By the way, I hate that the fourth instance of MemtestPro ALWAYS has to be force quit after stopping the test :devil:

Julian.

JBT27
06-01-2010, 01:43 AM
I hate to be a doom merchant - especially when it affects me :D - but this morning when I launched Modeler, the screen froze and was covered in an array of jagged fuzzy dots ..... that's the gpu failing ..... first time in a year I've had that.

I seriously doubt that doesn't have anything to do with all that we've been talking about here .....

Sigh.

Julian.

Oedo 808
06-01-2010, 06:47 AM
Oh man, you really are having a fun time with your system. I wouldn't worry about having 'done' anything, so long as you stick within the suggested ranges and you should be fine.

I can't remember, was the RAM purchased all at the same time? I would pursue the feedback on the OCZ forum and push the issue of not being able to get it stock stable. While having so many sticks be bad is unlikely, were they to be all from the same batch it's not so wild a possibility. Having read some of the comments from a couple of threads like this (http://forums.overclockers.co.uk/showthread.php?t=18144235) and this (http://forums.overclockers.co.uk/showthread.php?t=18147598) from the OcUK (http://forums.overclockers.co.uk/) forums does make me think 'hmm'. That you had auto on for the RAM settings is a pretty good indicator of why is wasn't stable before, but running with advertised volts and timings should have sorted that, you shouldn't have needed to change much, if anything else unless you were going for an overclock.

Strange that you were getting GPU glitches, with the PCI-e freq set to 100 you should have been fine there. A dodgy power-supply could affect this or maybe even a corrupt OS, but it could be just as many more things when you head down that road. Though if that's cheap ****** generic PSU I wouldn't be surprised.

As for the BIOS flashing, well it's not really anything that should give any cause for concern, it's just that in the unlikely event that it did go wrong, you'd have no way of accessing the BIOS to retry. My mobo (P45-DS4) has a backup BIOS that will auto re-flash if there are any issues like that, I don't know if yours has something similar.

It's a tough one to call, I was convincing myself that there was a faulty component somewhere on mine before I got my system stable, but then there is a good distance between stabilizing an overclock and getting it to run at stock without issue.

If I were in your shoes, try the replacement PSU, see if you can borrow a stick of RAM of someone and reload BIOS defaults, changing only the settings to match the RAM specs and take it from there.

RLM
06-01-2010, 06:50 AM
Hi Julian,

I sent you a pm.

RLM

JBT27
06-01-2010, 01:38 PM
Hi Julian,

I sent you a pm.

RLM

Thanks! Just responded!

Julian.

JBT27
06-01-2010, 01:44 PM
Oh man, you really are having a fun time with your system. I wouldn't worry about having 'done' anything, so long as you stick within the suggested ranges and you should be fine.

I can't remember, was the RAM purchased all at the same time? I would pursue the feedback on the OCZ forum and push the issue of not being able to get it stock stable. While having so many sticks be bad is unlikely, were they to be all from the same batch it's not so wild a possibility. Having read some of the comments from a couple of threads like this (http://forums.overclockers.co.uk/showthread.php?t=18144235) and this (http://forums.overclockers.co.uk/showthread.php?t=18147598) from the OcUK (http://forums.overclockers.co.uk/) forums does make me think 'hmm'. That you had auto on for the RAM settings is a pretty good indicator of why is wasn't stable before, but running with advertised volts and timings should have sorted that, you shouldn't have needed to change much, if anything else unless you were going for an overclock.

Strange that you were getting GPU glitches, with the PCI-e freq set to 100 you should have been fine there. A dodgy power-supply could affect this or maybe even a corrupt OS, but it could be just as many more things when you head down that road. Though if that's cheap ****** generic PSU I wouldn't be surprised.

As for the BIOS flashing, well it's not really anything that should give any cause for concern, it's just that in the unlikely event that it did go wrong, you'd have no way of accessing the BIOS to retry. My mobo (P45-DS4) has a backup BIOS that will auto re-flash if there are any issues like that, I don't know if yours has something similar.

It's a tough one to call, I was convincing myself that there was a faulty component somewhere on mine before I got my system stable, but then there is a good distance between stabilizing an overclock and getting it to run at stock without issue.

If I were in your shoes, try the replacement PSU, see if you can borrow a stick of RAM of someone and reload BIOS defaults, changing only the settings to match the RAM specs and take it from there.

Thanks! Sound advice on the PSU - I think I've done enough adjustments now, including using defaults, to show that something is seriously amiss between RAM and mobo, or perhaps as you suggest, the RAM is just plain dodgy ..... even though we've done all this work with it. Been working with the machine all day, and apart from that freaky GPU freakout, it's been fine - I don't really equate that with bad RAM. Very frustrating :)

Next time we get new boxes, I'm checking the RAM from the start!

I don't know if ours has a BIOS save/auto reinstate thing - it'd be very cool, but I'll bet it doesn't :) Thank god I'm naturally pessimistic with all this going on - even less RAM errors is something to celebrate!

I guess a stock box should just work from the start, where there is no OCing in place. And as I say, but for this odd locking-up problem, I probably wouldn't have stumbled on this RAM issue.

Julian.

RLM
06-01-2010, 02:27 PM
If I were in your shoes, try the replacement PSU, see if you can borrow a stick of RAM of someone

That's certainly not a bad idea, it's a process of elimination. If you can get that stable would be a better base to then consider flashing the bios from.

This is the QVL for your board

http://download.gigabyte.eu/FileList/Memory/motherboard_memory_ga-ep45-ud3.pdf

Just taking a couple of options from the 1066 range

http://www.crescentelectronics.co.uk/transcend-axeram-memory-2-gb-2-x-1-gb-dimm-240pin-ddr2.html

http://www.okobe.co.uk/ws/product/Kingston+KVR1066D2N71G+ValueRAM+Memory+Module+1GB+ DDR2+SDRAM+1066MHz/1000009896

RLM

JBT27
06-02-2010, 03:01 AM
That's certainly not a bad idea, it's a process of elimination. If you can get that stable would be a better base to then consider flashing the bios from.

This is the QVL for your board

http://download.gigabyte.eu/FileList/Memory/motherboard_memory_ga-ep45-ud3.pdf

Just taking a couple of options from the 1066 range

http://www.crescentelectronics.co.uk/transcend-axeram-memory-2-gb-2-x-1-gb-dimm-240pin-ddr2.html

http://www.okobe.co.uk/ws/product/Kingston+KVR1066D2N71G+ValueRAM+Memory+Module+1GB+ DDR2+SDRAM+1066MHz/1000009896

RLM

PSU first for sure - but I'm going to try to see about borrowing some RAM.

Thanks for the QVL - I had actually checked that already and had noted that our RAM is not on the list, though Gigabyte are covered by saying they can't test everything! Which is true of course. :)

Julian.